Huaizhong Zhu

Human-database interaction is inevitable in intelligent system applications, and accurately converting user-entered natural language into database query language is a critical step. To improve the accuracy, generalization, and robustness of text-to-SQL, we propose Nabil (a model for natural language conversion query language based on brain-inspired computing technology and a large language model). This model first leverages the spatiotemporal encoding capabilities of spiking neural networks to capture semantic features of natural language, then fuses these features with those generated by a large language model. Finally, a champion model is designed to select the optimal query from multiple candidate SQLs. Experiments were conducted on three database engines, DuckDB, MySQL, and PostgreSQL, and the model’s effectiveness was verified on benchmark datasets such as BIRD. The results show that Nabil outperforms existing baseline methods in both execution accuracy and effective efficiency scores. Furthermore, our proposed normalization and syntax tree abstraction algorithms further enhance the champion model’s discriminative capabilities, providing new insights for text-to-SQL research.

Cardiovascular disease is a typical chronic disease. The incidence of cardiovascular disease in the elderly population has grown to exceed 50%. The further aging of the current population has brought additional pressure on public health services. Therefore, to reduce the service pressure of medical institutions, accurate prediction of cardiovascular disease risk has become an essential task in intelligent elderly care. In order to achieve accurate prediction of cardiovascular disease risk, we propose a cardiovascular disease risk prediction model Acdim (a cardiovascular disease risk prediction model) based on ResNet (Residual Network), zebra optimization algorithm (ZOA), TabNet (Attentive Interpretable Tabular Learning), and AdaBoost (Adaptive Boosting) algorithms. In training the Acdim model, we used the dat set of the World Health Organization (WHO) Behavioral Risk Factor Monitoring System records. In the experiment, the Acdim model achieved an accuracy of 96%, a precision of 94%, a recall of 93%, a specificity of 95%, an F1 score of 91%, and an AUC of 95%. The experimental results show that the proposed Acdim model can enable elderly care institutions to accurately predict the risk of cardiovascular disease in the absence of doctors.

Gesture recognition is widely used in communication, virtual reality, smart cockpit, smart home, etc. This paper proposes an application system based on MediaPipe for intelligent recognition of 26 letter gestures in English. The system uses images captured by a monocular camera, reads images using OpenCV, creates a hand detection model, uses MediaPipe to identify 21 key points of the hand, constructs a gesture recognition module for English letters, and recognizes letters through rough gesture recognition algorithm and fine gesture recognition algorithm. The experimental results show that the recognition accuracy of all English letters in this system is more than 81%, and the recall rate is more than 83%, which can be applied to gesture recognition related applications of English letters.

Students’ actual learning engagement in class, which we call learning attention, is a major indicator used to measure learning outcomes. Obtaining and analyzing students’ attention accurately in offline classes is important empirical research that can improve teachers’ teaching methods. This paper proposes a method to obtain and measure students’ attention in class by applying a variety of deep-learning models and initiatively divides a whole class into a series of time durations, which are categorized into four states: lecturing, interaction, practice, and transcription. After video and audio information is taken with Internet of Things (IoT) technology in class, Retinaface and the Vision Transformer (ViT) model is used to detect faces and extract students’ head-pose parameters. Automatic speech recognition (ASR) models are used to divide a class into a series of four states. Combining the class-state sequence and each student’s head-pose parameters, the learning attention of each student can be accurately calculated. Finally, individual and statistical learning attention analyses are conducted that can help teachers to improve their teaching methods. This method shows potential application value and can be deployed in schools and applied in different smart education programs.

As the core component of permanent magnet motor, the magnetic tile defects seriously affect the quality of industrial motor. Automatic recognition of the surface defects of the magnetic tile is a difficult job since the patterns of the defects are complex and diverse. The existing defect recognition methods result in difficulty in practical application due to the complicated system structure and the low accuracy of the image segmentation and the target detection for the diversity of the defect patterns. A self-supervised learning (SSL) method, which benefits from its nonlinear feature extraction performance, is proposed in this study to improve the existing approaches. We proposed an efficient multihead self-attention method, which can automatically locate single or multiple defect areas of magnetic tile and extract features of the magnetic tile defects. We also designed an accurate full-connection classifier, which can accurately classify different defects of magnetic tile defects. A knowledge distillation process without labeling is proposed, which simplifies the self-supervised training process. The process of our method is as follows. A feature extraction model consists of standard vision transformer (ViT) backbone, which is trained by contrast learning without labeled dataset that is used to extract global and local features from the input magnetic tile images. Then, we use a full-connection neural network, which is trained by using labeled dataset to classify the known defect types. Finally, we combined the feature extraction model and defect classification model together to form a relatively simple integrated system. The public magnetic tile surface defect dataset, which holds 5 defect categories and 1 nondefect category, is used in the process of training, validating, and testing. We also use online data augmentation techs to increase training samples to make the model converge and achieve high classification accuracy. The experimental results show that the features extracted by the SSL method can get richer and more detailed features than the supervised learning model gets. The composite model reaches to a high testing accuracy of 98.3%, and gains relatively strong robustness and good generalization ability.

A heterogeneous traffic flow consists of regular vehicles, and intelligent connected vehicles having interactive functions is updating the composition of the current urban-road network traffic flow. It has been a growing trend and will continue to be so. Because of the urgent demand, the research focused on three main parts of cooperative control methods under intelligent connected vehicles environment, typical traffic control application scenarios and experimental validation in intelligent connected vehicles conditions, and intersection-oriented hybrid traffic control mechanism for urban road. For heterogeneous interrupted traffic flow of intelligent connected vehicles, to analyze the characteristics and information extraction method of heterogeneous traffic flow of intelligent connected vehicles under different conditions, the research examined driving modes of regular vehicles and intelligent connected vehicles, including car following and lane changing. This study summarized control modes of traffic-signal control, active control of intelligent connected vehicles, and indirect control of regular vehicles through intelligent vehicles to study the active control mechanism and multi-intersection coordinated control strategy for intelligent connected vehicle heterogeneous traffic flow. With the combination of coordinated control theory, this work overviewed integrated experiment of information interaction and coordinated control under intelligent-connected-vehicle heterogeneous traffic-flow environments.

This paper explores difference between Advise and Suggest in the fields of collocation, semantic prosody and register. Under the aid of the British National Corpus (BNC), the author collects and samples evidence accordingly, finds out great difference between them, and points out some false friends we make everyday. Analyzed in a more scientific way, this paper aims to provide a corpus-based method for readers to perceive English words.

The prediction of entrance and exit passenger flow of rail transit stations is one of key research focuses in the area of intelligent transportation. Based on the big data of rail transit IC card (Public Transportation Card), this paper analyzes the data of major dynamic factors having effect on entrance passenger flow and exit passenger flow of rail transit stations: weather data, atmospheric temperature data, holiday and festival data, ground index data, and elevated road data and calculates the daily entrance passenger flow and daily exit passenger flow of individual rail transit stations with data reduction. Furthermore, based on the history data of passenger flow of rail transit stations and relevant influence factors, it applies the deep learning method to choose the relatively optimal hidden layer node by means of the cut-and-try method, set up input data and labeled data, select the activation function and loss function, and use the Adam Gradient Descent Optimization Algorithm for iterative global convergence. The results verify that this method accurately predicts the daily entrance passenger flow and daily exit passenger flow of rail transit stations with the prediction error of less than 4.1%. Finally, the proposed model is compared with the linear regression model.